| In recent years, the development of novel techniques for genome sequencing and other high-throughput methods has enabled the identification and quantification of individual cell components. Genome-scale metabolic models (GSMMs) have been developed for several organisms, including humans. Under the framework of constraint-based modeling, these have provided phenotype prediction methods, useful in fields as metabolic engineering and biomedical research, spanning tasks as drug discovery, biomarker identification and host-pathogen interactions, and targeting diseases such as cancer, Alzheimer, or diabetes. However, these methods have been limited, since the human body has a diversity of cell types and tissues making the development of specific models an imperative. Methods to provide phenotype simulation with the integration of omics data and to automatically generate tissue-specific models, based on generic human metabolic models and a plethora of omics data, have been proposed. However, their results have not been adequately and critically evaluated and compared. Moreover, their usage is restricted to users with computer science skills, since they are not available in user-friendly software platforms. In this work, an open-source software framework for the integration of GSMMs with omics data has been provided. It contains methods for the processing and integration of data with models, for the reconstruction of tissue-specific GSMMs and for phenotype simulation using omics data. A user-friendly graphical interface is provided for non-programming users to be able to run these methods, while an open programming interface allows the community to contribute. The methods have also been validated and compared in representative case studies, being studied the effects of data sources and algorithms in the final results. In particular, glioblastoma has been selected as a more comprehensive case study, where specific models were generated for a representative cell line using different approaches. These have been compared and integrated into a consensus model, which has been further used for analysis and to support phenotype simulation. The results allow insights into cancer metabolism and possible routes towards drug discovery. |
